Is QBism Bayesian?
I've been exploring QBism, an interpretation of Quantum Mechanics (QM) based on or inspired by Bayesian inference. The appeal being that mysteries about the ontic or epistemic nature of the wavefunction and about the collapse of the wavefunction could be resolved in this perspective.
I found it confusing and incoherent. There is nothing mathematically Bayesian about it. We don't describe beliefs with probability theory alone, as we use a wavefunction, and we don't update them using Bayes' theorem alone, as we need the Born rule. It is, though, Bayesian in spirit, in that the wavefunction is interpreted personally and epistemically: it isn't real or ontic, it something that we create in our heads to describe a quantum state. The collapse of a wavefunction isn't real either, it's just an update we do in our heads, analogous to Bayesian updating from prior to posterior. In later works this is explicitly stated and the B is said to stand for Bruno de Finetti rather than Bayes, as much like de Finetti said that probabilities didn't exist, QBism says wavefunctions don't exist in a physical, ontic way.
Sounds wonderful at first glance, until you ask a few more questions. Why are we using wavefunctions to describe our epistemic state (cf. probabilities)? Why do we use the Born rule to associate probabilities with outcomes? If the wavefunction is epistemic, what is it that we are ignorant about? If the measurement and update are all in our heads, what determines which random outcome we observe?
I found the answers to these questions vague and handwavy, as they involve 'agents', 'agent experiences', and 'normative rules'.
Tags: quantum-mechanics, bayes